1. Introduction

Although oral antidepressants (ADs) are important treatment modalities for managing depression, over half of patients do not respond to the first course of AD, and a significant proportion remain nonresponsive to subsequent lines of treatment [1]. This condition is “treatment-resistant depression (TRD)”; a source of immense burden in mortality, morbidity, and economic cost associated with patients with depression. Patients with TRD have a 29%–39% higher risk of all-cause mortality compared to those without TRD. In addition to the elevated risk of self-harm, TRD is also associated with the development of various physical comorbidities—particularly cardiovascular disease and stroke—which further increase mortality. Moreover, TRD is closely linked to higher direct medical costs, productivity losses, and employment changes [2,3]. Pharmaceutical strategies to manage TRD commonly include switching between oral ADs, combining oral ADs, or augmenting current therapy with antipsychotics [4,5]. However, they are unlikely to improve outcome if patients are pharmacologically nonresponsive to the existing drug classes within oral ADs.

Esketamine nasal spray was first approved by the U.S. Food and Drug Administration (FDA) and the European Medicines Agency (EMA) in 2019 for treating adults with TRD in combination with an oral AD [6,7]. As a novel antidepressant with rapid onset and a distinct mechanism of action, esketamine has received considerable policy interest and clinical attention internationally. To date, esketamine has been approved in 77 countries and used by over 140,000 patients worldwide [8]. Distinct from conventional oral ADs, esketamine is an N-methyl-D-aspartate receptor antagonist that mainly upregulates the release of brain-derived neurotrophic factor, a crucial substance restoring synaptic remodeling which takes weeks to produce with oral ADs [9,10]. The intranasal route of administration further contributes to its rapid onset of symptom relief. Compared with oral AD monotherapy, Phase III trials show that esketamine in addition to an oral AD significantly reduced symptoms of depression (measured using the Montgomery–Åsberg Depression Rating Scale (MADRS) score) after 4 weeks, with immediate change as early as the second day of treatment which continued for 28 days, in contrast to 2–4 weeks with oral ADs [11]. The same trials found that continued treatment with esketamine among responding patients prolonged the median time until relapse by 7.5 weeks among the full population in each arm with a relapse rate of 26.7% (45.3% in the control) [12].

Although efficacy is favorable, esketamine currently has a high market price. Ross and colleagues found that in the United States esketamine as adjunct to oral AD compared to oral AD alone was unlikely to be cost-effective, unless the price fell by 40% [13]. Five other modeling studies in the United States, Italy, and Canada gave similar conclusions when comparing esketamine to oral AD, intravenous ketamine, and electroconvulsive therapy (ECT) [14–18]. Given the reported price up to US$5,000 per month [13], the health and economic benefits could be reasonably outweighed by the additional cost. Unlike most pharmacological treatments, patients must be supervised by healthcare professionals during drug administration, which further raises the cost for clinical monitoring.

However, the existing cost-effectiveness studies had limitations. Most models compared adjunctive esketamine to oral AD alone, but the latter is not a realistic comparator. Based on clinical guidelines, treatment pathway research and technical appraisal documents by the National Institute for Health and Care Excellence (NICE) in the United Kingdom, oral AD monotherapy was not the mainstay of treatment among depression patients who had already failed two lines of treatment [2,4,5,19]. Of all real-world drug treatments for TRD in the Hong Kong setting, oral AD monotherapy accounts for only 42% of choices, as patients tend to have combination/augmentation therapies (42%) or polytherapy (16%) [2]. Nonpharmaceutical choices such as psychotherapy and neurostimulation therapies (e.g., rTMS and ECT) are also common interventions for TRD. One reason for solely comparing to oral AD monotherapy was the heavy reliance on Phase III esketamine trial data, which did not apply the usual real-world treatments for TRD as the control. Options other than oral AD monotherapy were therefore modeled inadequately but still of economic interest.

Motivated by the clinical and policy interest in esketamine as a novel and rapid-acting antidepressant, this study aims to compare esketamine nasal spray with six commonly used real-world treatment options to provide a comprehensive economic evaluation of third-line strategies for TRD management from the healthcare payer’s perspective in Hong Kong. In accordance with the NICE technology appraisal discussion (Sections 3.3 and 3.6) [19], we chose augmentation therapy as the reference comparator. In real-world clinical practice, patients with TRD are typically offered augmentation therapy following two unsuccessful antidepressant treatments. By systematically evaluating multiple real-world comparators, this study not only assesses the relative economic value of esketamine but also seeks to identify clinically relevant and cost-effective treatment strategies.

2. Methods

This study was conducted in accordance with the Consolidated Health Economic Evaluation Reporting Standards (CHEERS 2022) checklist [20]. The study protocol was approved by the Institutional Review Board of the University of Hong Kong/Hospital Authority Hong Kong West Cluster (HKU/HA HKW IRB) (Approval No.: UW 20-218). The requirement for informed consent was waived due to the use of fully anonymized data. All model inputs, assumptions, and results were cross-checked by two independent researchers (YL and VKYC) to ensure quality, accuracy, and transparency.

2.3. Model description

We adapted a published state-transition model commonly used for economic evaluations for antidepressants (Fig 1) [13,17]. The model consists of at least three treatment lines with five health states per line, including: Treatment, Response, Remission, Nonresponse, and Relapse. Upon ascertainment as TRD, patients were initiated on the first treatment (equivalent to the third-line therapy since depression diagnosis) as specified in Section 2.2 for 4 weeks before proceeding to other states in the next cycle based on clinical improvement. “Response” frequently refers to at least a 50% reduction in depression symptom scores, whilst “remission” refers to achieving a score below an absolute threshold originally defined by the sourced articles which may differ between the choice of measuring instruments. “Nonresponse” commonly refers to a state not responding nor remitting to treatment, and “relapse” is the recurrence of symptoms after response and remission when the absolute score exceeds a threshold pre-defined in the scales. When patients experienced a relapse or nonresponse, they would initiate a next-line treatment in the ensuing cycle. Lastly, all patients in any states may proceed to all-cause mortality (absorbing state).

Download:

• PNG
  larger image
• TIFF
  original image

Fig 1. Schema of Markov model structure.

https://doi.org/10.1371/journal.pmed.1005047.g001

3. Results

3.1. Base-case analysis

Table 1 presents the results of the base-case cost-effectiveness analysis per patient. Under a WTP threshold of US$50,000/QALY, esketamine was not cost-effective when compared to augmentation therapy, combination therapy, psychotherapy alone, and psychotherapy plus AD, with incremental cost-effectiveness ratios (ICERs) ranging from US$134,127 to US$312,750 per QALY. In contrast, esketamine dominated rTMS, being both less costly and more effective. Compared with esketamine, ECT was associated with a high ICER (US$322,407/QALY), exceeding the WTP threshold. Combination therapy was the most cost-effective strategy, with all other treatment options having higher ICERs when compared against the reference group (AUG).

Download:

• PNG
  larger image
• TIFF
  original image

Table 1. Base-case analysis of per patient cost, QALYs and ICER in each comparative arm.

https://doi.org/10.1371/journal.pmed.1005047.t001

The solid gray line in Fig 2 connects the nondominated strategies which forms the efficiency frontier. Psychotherapy alone and rTMS adjunctive to an oral AD fall above this frontier, an indication that they are dominated strategies, being more costly and less effective than at least one alternative. On the cost-effectiveness frontier, combination therapy was the least costly nondominated strategy, followed by psychotherapy plus antidepressant therapy (PSY + AD), esketamine plus antidepressant therapy (ESK + AD), and electroconvulsive therapy plus antidepressant therapy (ECT + AD). The ICER for ESK + AD compared with the next least costly nondominated alternative (PSY + AD) was US$312,750 per QALY. Moving further along the frontier, the ICER for ECT + AD compared with ESK + AD was US$322,407 per QALY.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 2. Cost-effectiveness efficiency frontier of treatment-resistant depression (TRD) strategies.

Each point represents a third-line treatment strategy for TRD compared with augmentation therapy. Red dashed lines indicate willingness-to-pay (WTP) thresholds of US$50,000 and US$150,000 per quality-adjusted life-year (QALY). The solid gray line represents the efficiency frontier formed by nondominated strategies. Gray dashed lines represent indifference curves. Strategies located below and to the left of a curve are considered more cost-effective than those above and to the right. Abbreviations: WTP, willingness-to-pay; QALY, quality-adjusted life-year; AUG, Augmentation therapy (antidepressant combined with antipsychotic/lithium); COM, Combination therapy (antidepressant combined with antidepressant); PSY alone, Psychotherapy alone; PSY + AD, Psychotherapy combined with antidepressant; ESK + AD, Esketamine combined with antidepressant; rTMS + AD, Repetitive transcranial magnetic stimulation combined with antidepressant; ECT + AD, Electroconvulsive therapy combined with antidepressant.

https://doi.org/10.1371/journal.pmed.1005047.g002

3.3. Sensitivity analyses

The top 10 influential parameters for each comparison are displayed in tornado diagrams for the DSA (Fig B in S1 Appendix). Across all comparisons, model parameters related to the efficacy of esketamine—particularly the relative risks (RRs) of transitioning from remission to relapse and from treatment to remission—exerted the greatest influence on ICER values. In the comparison between esketamine and augmentation therapy (reference group), varying the RR of remission to relapse for the esketamine strategy led to ICER fall below three times GDP (US$150,000/QALY). Despite this wide variation, the cost-effectiveness conclusion remains unchanged.

PSA results in Fig 3 show that while most PSA simulations for esketamine exceed the US$50,000/QALY threshold, its cost-effectiveness is generally more favorable than ECT and rTMS. These findings are consistent with the CEACs (Fig 4). At a WTP of US$50,000/QALY, combination therapy had the highest probability of cost-effectiveness (75%), followed by augmentation therapy (25%), while esketamine was not cost-effective (0%). Even at US$150,000/QALY, esketamine remained less likely to be cost-effective (15.1%) compared to combination (58.3%) and augmentation therapy (15.6%). Esketamine only surpassed other strategies in cost-effectiveness probability when the WTP threshold exceeded US$213,675/QALY—more than four times the base-case threshold.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 3. Probabilistic sensitivity analysis results (10,000 Monte Carlo simulations) on the incremental cost-effectiveness scatter plot.

Each dot represents one simulation. Ellipses indicate 95% uncertainty intervals. The dashed lines represent willingness-to-pay (WTP) thresholds of US$50,000 and US$150,000 per quality-adjusted life-year (QALY). Abbreviations: AUG, augmentation therapy; COM, antidepressant combination therapy; cPSY, psychotherapy alone; mPSY, psychotherapy combined with antidepressant; ESK, esketamine combined with antidepressant; TMS, repetitive transcranial magnetic stimulation combined with antidepressant; ECT, electroconvulsive therapy combined with antidepressant; WTP, willingness-to-pay; QALY, quality-adjusted life-year.

https://doi.org/10.1371/journal.pmed.1005047.g003

Download:

• PNG
  larger image
• TIFF
  original image

Fig 4. Cost-effectiveness acceptability curves (CEACs) based on 10,000 Monte Carlo simulations.

Curves show the probability that each treatment strategy is the most cost-effective across a range of willingness-to-pay (WTP) thresholds. Gray dashed lines indicate the base-case WTP threshold of US$50,000 per quality-adjusted life-year (QALY) and the three times GDP per capita threshold of US$150,000 per QALY. The threshold at which esketamine first becomes the most cost-effective option is US$213,675 per QALY. Abbreviations: CEAC, cost-effectiveness acceptability curve; WTP, willingness-to-pay; QALY, quality-adjusted life-year; AUG, augmentation therapy; COM, antidepressant combination therapy; cPSY, psychotherapy alone; mPSY, psychotherapy combined with antidepressant; ESK, esketamine combined with antidepressant; TMS, repetitive transcranial magnetic stimulation combined with antidepressant; ECT, electroconvulsive therapy combined with antidepressant.

https://doi.org/10.1371/journal.pmed.1005047.g004

4. Discussion

In this study, we compared esketamine nasal spray with six commonly used third-line treatment strategies for TRD and found that its economic value varied substantially across different comparators. Unlike previous studies that primarily compared esketamine with oral AD monotherapy or only single comparator, our model incorporates a broader spectrum of both pharmacological and nonpharmacological comparators, better reflecting actual clinical practice. Beyond assessing the value of esketamine, our analysis also identifies the most cost-effective treatment option relative to the realistic multiple alternative comparator therapies. Notably, our study improves upon earlier models by leveraging a territory-wide real-world database, rather than relying solely on international trial-based evidence. This design enhances both the contextual relevance and the generalizability of our findings to routine clinical settings.

The base-case results indicated that esketamine was not cost-effective compared to most other treatment strategies under the WTP threshold of US$50,000 per QALY, except when compared with rTMS and ECT. Among all therapies evaluated, combination therapy emerged as the most cost-effective option. Our results are broadly consistent with prior economic evaluations of esketamine conducted in the United States, Canada, and Europe. For example, Ross and colleagues reported base-case ICERs of approximately US$237,000–242,000/QALY from both societal and healthcare sector perspectives in the United States, suggesting that esketamine is unlikely to be cost-effective at conventional thresholds. They also indicated that esketamine would require substantial price reductions (>40%) to reach acceptable cost-effectiveness [13]. Similar conclusions were drawn by Canada’s CADTH report, where esketamine’s ICER was approximately US$125,376/QALY, necessitating an estimated 60% price reduction to achieve cost-effectiveness at a WTP of US$50,000/QALY [18]. Likewise, Rognoni and colleagues found esketamine was not cost-effective from the Italian healthcare system perspective but could be justified from a societal perspective, primarily due to productivity gains from rapid functional improvement and reduced work loss [17].

However, our finding that esketamine appears more cost-effective than ECT contrasts with the previous economic evaluation in which ECT was found to be cost-effective or even dominant over esketamine [16]. This discrepancy is not unexpected and can be largely explained by two key methodological differences: cost structure assumptions and effectiveness modeling choices.

First, substantial differences exist in how ECT costs are defined across studies. In our analysis, the unit cost of ECT was based on the Hospital Authority public charge for noneligible persons in Hong Kong, reflecting the full, unsubsidized procedural cost, including anesthesiology, psychiatric consultation, nursing labor, recovery facilities, and institutional overheads. In contrast, this previous study derived ECT costs from NICE-based bundled tariffs (approximately £558 per session), which represent subsidized average payments within the UK National Health Service rather than disaggregated, labor-intensive costs. Given that ECT is predominantly labor-driven, such differences in healthcare financing and labor cost structures can substantially influence total cost estimates and downstream cost-effectiveness results. This interpretation is further supported by our DSA (Fig B in S1 Appendix). When ECT-related procedural and inpatient admission costs were varied by ±50%, the incremental cost-effectiveness ratio for ECT versus esketamine changed substantially, confirming that comparative cost-effectiveness results are highly sensitive to labor- and procedure-related cost assumptions. Although reductions in ECT costs markedly lowered the ICER, ECT did not become cost-effective relative to esketamine at the base-case WTP threshold of US$50,000/QALY in any of the scenarios examined. Together, these findings underscore that differences in healthcare financing structures and labor cost accounting can materially affect cost-effectiveness estimates for labor-intensive interventions such as ECT, and highlight the importance of context-specific costing in economic evaluations.

Second, differences in modeled effectiveness assumptions may further contribute to divergent findings. In our model, ECT efficacy parameters were primarily sourced from the broader TRD literature and were not consistently restricted to the most severe or urgent clinical subgroups for whom ECT is typically indicated in real-world practice. As a result, the incremental QALY gains associated with ECT may be conservatively estimated. If ECT effectiveness were higher among patients with more severe, refractory, or life-threatening depression as suggested in some clinical studies, its true QALY benefit could exceed that captured in our base-case analysis, thereby improving its cost-effectiveness profile. This interpretation is supported by our DSA (Fig B in S1 Appendix). When the relapse risk from remission under ECT was reduced (relative risk of remission-to-relapse = 0.45), the ICER for esketamine versus ECT decreased substantially to approximately US$191,000 per QALY, approaching three times the conventional WTP threshold. This scenario reflects a clinical context in which ECT is more effective at sustaining remission, as may be expected among more severe or highly refractory patient populations. Although ECT did not become cost-effective relative to esketamine even under this more optimistic effectiveness assumption, the marked reduction in the ICER illustrates that ECT cost-effectiveness is highly sensitive to assumptions regarding relapse prevention and patient severity. Therefore, the combination of higher modeled ECT costs and potentially conservative assumptions regarding ECT effectiveness in our analysis helps explain why esketamine appears economically favorable in the Hong Kong–specific setting, whereas studies conducted in lower-labor-cost or differently subsidized healthcare systems have reported contrasting conclusions.

Taken together, these considerations also have important implications for the generalizability of our findings. Our findings should be interpreted in the context of a high-income healthcare system with relatively high labor costs. Although esketamine appeared economically favorable compared with rTMS and ECT in Hong Kong, this result is largely driven by the labor-intensive nature of rTMS and ECT, which require substantial physician, nursing, anesthetic, and facility resources. In healthcare systems with lower-labor-costs, the relative cost-effectiveness of these interventions may differ substantially, even if clinical effectiveness remains unchanged. This is particularly relevant for other labor-intensive or service-dependent interventions. For example, treatments such as intravenous ketamine (IV ketamine) have cost structures that are highly dependent on administration, monitoring, and service delivery requirements, making their economic value sensitive to local implementation practices. The component-level cost reporting in this study enables context-specific re-estimation and supports adaptation of the model to alternative healthcare settings.

Moreover, sensitivity and scenario analyses did not materially alter the conclusions of the base-case results. DSA identified the efficacy parameters of esketamine—particularly the relative risks of transitioning from remission to relapse—as the most influential drivers of cost-effectiveness. However, even under the most favorable assumptions, the resulting ICER still remained above the WTP threshold. PSA further confirmed the limited cost-effectiveness of esketamine, with a 0% probability at the WTP threshold, and only a 15.1% probability at a threshold three times the WTP threshold (US$150,000/QALY). Esketamine was only observed to be the most economically favorable strategy when the WTP threshold exceeded approximately US$213,675/QALY. In Scenario 6, which examined reductions in the esketamine drug acquisition price alone, we found that a 75% price reduction was sufficient to reduce the ICER to below a higher WTP threshold corresponding to three times WTP (US$150,000/QALY), but remained above the conventional WTP (US$50,000/QALY). By contrast, achieving cost-effectiveness at the WTP threshold would require a substantially larger reduction in total esketamine arm costs, rather than drug price alone. Therefore, in Scenario 8, under highly optimistic structural assumptions—including linear drug pricing and a clear separation of drug and monitoring cost components—a reduction of ~35% in total esketamine-related costs would be required to approach a WTP threshold of US$50,000/QALY. At higher thresholds, the probability of cost-effectiveness increases, as shown in the CEACs (Fig 4).

While none of the esketamine scenarios achieved cost-effectiveness under the current threshold, scenario analysis offered valuable insights into potential strategies to improve its economic profile. Notably, Scenario 2, which extended the model cycle length from 4 to 8 weeks to reflect a longer clinical follow-up interval, resulted in a substantially lower ICER (US$73,556/QALY). Importantly, this improvement was driven by structural features of the model. Specifically, fewer transition opportunities per year and prolonged time spent in response and remission states, rather than by any assumed increase in treatment efficacy. Therefore, this scenario should be interpreted as an exploratory, upper-bound structural scenario rather than as evidence that less frequent clinical monitoring itself improves health outcomes. Although existing clinical guidelines often recommend follow-up intervals of one to two months in stable patients, the findings from Scenario 2 do not imply a causal benefit of reduced monitoring frequency, but instead illustrate how cost-effectiveness estimates may be sensitive to assumptions regarding cycle length and transition timing. Additionally, substituting specialist outpatient clinic and psychiatric day hospital services with simplified nurse-led monitoring and intranasal protocols substantially reduced overall costs. These results imply that apart from considerations around drug pricing, policy interventions aimed at optimizing care delivery may also improve the cost-effectiveness of esketamine to some extent.

Despite esketamine’s current limited economic appeal, it remains a potentially valuable option due to the rapid onset of action and relapse prevention benefits for patients with severe TRD who have not responded to other therapies. However, routine use may not be economically viable at existing price and reimbursement levels. In contrast, our findings importantly highlight that combination therapy emerged as the most cost-effective third-line strategy, yet it remains underutilized in real-world clinical practice. More broadly, our results suggest that future economic evaluations may benefit from more focused comparisons between clinically similar third-line interventions, such as esketamine and rTMS. Because these treatments are often used in overlapping patient populations but differ fundamentally in cost structure (drug-intensive versus labor-intensive), such comparisons may yield more transferable and methodologically robust insights across healthcare systems. Policymakers could consider increasing awareness and promoting wider adoption of combination therapy, as this would likely yield substantial economic and clinical benefits. For esketamine, strategies such as price negotiations or conditional reimbursement schemes could enhance its cost-effectiveness, given the substantial quality of life improvements associated with its use. Additionally, health authorities could also consider applying higher WTP thresholds to economic evaluations. More broadly, findings from comparative cost-effectiveness analyses may help inform more comprehensive and context-sensitive treatment sequencing strategies for TRD, beyond conventional line-of-therapy frameworks. Such evidence may support clinical decision-making and resource allocation in later-line treatment settings.

This study has several limitations. First, in the absence of head-to-head trials directly comparing esketamine with other real-world treatment strategies, our analysis relied on anchored indirect comparisons using antidepressant monotherapy as a common reference. While this approach allows internally consistent comparisons across treatment arms within the model framework, the effectiveness estimates were drawn from clinical trials and meta-analyses conducted in populations that may differ in baseline disease severity, degree of treatment resistance, and definitions of response or remission. As a result, the comparative effectiveness estimates may be subject to residual confounding arising from heterogeneity across evidence sources and should be interpreted with caution. In particular, some interventions, such as ECT, are typically reserved for a highly selected subgroup of patients with severe or refractory depression, applying average treatment effects across the broader TRD population may underestimate its health benefits for appropriate candidates. Consequently, the cost-effectiveness of ECT observed in this study should be interpreted in the context of population-level comparisons rather than severity-specific treatment pathways. Second, the limited availability of relapse and long-term response data for some comparators necessitated making assumptions based on similar treatment classes or extrapolations from the STAR*D study, which may not fully capture real-world treatment dynamics. Third, the model assumed constant treatment effects over time and did not account for potential variations in adherence, discontinuation, or dose adjustments that commonly occur in clinical practice. Moreover, we assumed that within the same Markov health state patients experienced the same health-related quality of life regardless of which treatment led to that state. While this approach is consistent with the way utility values are typically reported and applied in the major depressive disorder literature, it does not capture treatment-specific side effects or treatment burden. In reality, cognitive adverse effects associated with ECT, dissociative symptoms related to esketamine, or the frequent clinic visits required for rTMS may differentially affect patients’ quality of life beyond depression symptom status alone. Ignoring such treatment-specific utility decrements may bias QALY estimates in either direction. Future studies incorporating treatment-specific utilities or patient-reported outcome measures would allow more refined comparisons across treatment modalities. Another limitation is that we did not stratify patients based on the degree of prior treatment resistance, which could influence treatment outcomes and potentially overestimate esketamine’s effectiveness among more severely resistant cases. Furthermore, our model did not account for sequential treatment scenarios involving switching or progressing through multiple treatment options, particularly the clinical scenario in which esketamine might be used as an effective option after other therapies have failed. Future studies could consider sequential treatment modeling, exploring the potential role of esketamine as a last-resort option following the failure of other more cost-effective therapies, thus better reflecting real-world clinical decision-making. Lastly, IV ketamine was not included as a comparator in this study. Although it is increasingly used in some healthcare settings, it is not routinely incorporated into standardized treatment pathways for TRD within the Hong Kong public healthcare system, and there is limited availability of locally relevant data on its utilization, costs, and real-world effectiveness. While it would be possible to approximate cost inputs using data from other settings, such assumptions would introduce substantial uncertainty and reduce the contextual validity of the model. In addition, its off-label use and heterogeneity in administration protocols present further challenges for robust economic modeling. Given its distinct cost structure—characterized by low drug acquisition costs but relatively higher administration and service-related costs—its cost-effectiveness is likely to be context-dependent and may vary substantially across healthcare systems. Future research incorporating IV ketamine as a comparator would be valuable in healthcare systems where its use is more established in clinical practice, enabling more context-specific and policy-relevant economic evaluations.

Overall, this study provides a comprehensive real-world–informed economic evaluation of esketamine compared with multiple third-line treatment strategies for TRD. While esketamine was not cost-effective compared with most alternatives under the current WTP threshold, it may offer relative economic advantages over certain neurostimulation treatments such as rTMS and ECT. Importantly, our findings highlight that cost-effectiveness evidence may help inform more context-sensitive treatment sequencing strategies beyond conventional line-of-therapy frameworks. These insights may support reimbursement decision-making, pricing negotiations, and the design of more efficient care delivery models for TRD management.

Supporting information

Acknowledgments

The authors would like to thank Zonglin Dai and Deliang Yang for their technical support in model implementation and data processing. We also thank Lisa Lam for her assistance with language editing and proofreading of the manuscript.